Odds ratios (with 95% confidence interval) for severe retinopathy of prematurity (ROP) (ROP stage ≥3) in relation to gestational age at birth and birth weight (univariate and multivariate logistic regression analyses).
Austeng D, Källen KBM, Ewald UW, Jakobsson PG, Holmström GE. Incidence of Retinopathy of Prematurity in Infants Born Before 27 Weeks' Gestation in Sweden. Arch Ophthalmol. 2009;127(10):1315-1319. doi:10.1001/archophthalmol.2009.244
To determine the incidence of retinopathy of prematurity (ROP) in extremely preterm infants born before 27 weeks' gestation in Sweden during a 3-year period.
A national, prospective, population-based study was performed in Sweden from April 1, 2004, to March 31, 2007. The ophthalmologic part of the study was separately organized, and screening for ROP was performed beginning postnatal week 5. The criteria for the treatment of ROP agreed with the recommendations of the Early Treatment for Retinopathy of Prematurity Cooperative Group.
During the study, 506 of 707 live-born infants survived until the first eye examination. Of these, 368 (72.7%) had ROP: 37.9% had mild ROP and 34.8% had severe ROP. Ninety-nine infants (19.6%) were treated. Gestational age at birth was a stronger predictor of ROP than was birth weight. A log-linear relationship between severe ROP and gestational age at birth was found in the present cohort, and the risk of ROP was reduced by 50% for each week of increase in gestational age at birth.
Today, extremely preterm infants are surviving, and this population-based study with ROP as a primary outcome shows a higher incidence of this condition than in previously reported national cohorts.
Retinopathy of prematurity (ROP) remains an important cause of childhood blindness and visual impairment throughout the world.1 Consecutive population-based studies2,3 on the incidence of ROP have been conducted in Stockholm County, Sweden. During the last decade, neonatal care has changed with an increase in centralization, implementation of new therapies, and provision of intensive care for infants of extremely low gestational age (GA). These changes have contributed to an increasing population of survivors in neonatal intensive care units today.4- 7 The incidence of ROP in these extremely premature infants is, therefore, unknown.
The main aim of the present study is to evaluate various aspects of ROP in a population of infants born before 27 weeks' gestation. This article focuses on the incidence of ROP and its relation to GA at birth. Subsequent articles will report the natural history of ROP in the survivors and discuss the role of birth weight (BW), growth, and other risk factors on the etiology of ROP and aspects of screening routines and treatment.
This study is part of a national project initiated by the Swedish Association of Perinatology and the Swedish National Board of Health and Social Welfare.8 The general aim of the project was to determine morbidity and mortality in extremely premature infants with a GA of less than 27 weeks born in Sweden during a 3-year period (April 1, 2004, to March 31, 2007). National data in Sweden and its various regions were available for all infants born with a GA of less than 27 weeks. Gestational age at birth was calculated using the expected date of delivery based on the results of an ultrasound examination performed before 20 weeks' gestation.
A group of pediatric ophthalmologists in charge of screening for ROP in their hospitals across Sweden was assembled before onset of the study. They were informed about the study procedures and protocols, and they regularly discussed diagnoses during the study, the stages of ROP, the definitions of plus disease and aggressive posterior ROP, and indications for treatment. Information and advice to participating ophthalmologists were also given at annual meetings of the Swedish Ophthalmologic Society and the Swedish Society of Medicine and at regional ophthalmologic meetings.
Screening for ROP included examinations beginning the fifth postnatal week and continuing until the retina was completely vascularized. The study protocol advocated weekly examinations, enabling study of the course and severity of ROP. In infants with no ROP or with mild ROP, that is stages 1 and 2, without progression during the latest examinations, further screening was performed every week or every other week from a postmenstrual age of 35 weeks until the retina was entirely vascularized or until regression of ROP. We used the International Classification of Retinopathy of Prematurity revisited to categorize ROP,9 and we followed the Early Treatment for Retinopathy of Prematurity Cooperative Group recommendations concerning criteria for treatment.10 In the present study of the incidence of ROP, each infant was classified according to the maximum stage of ROP in either eye.
The pupils were dilated twice with combined cyclopentolate hydrochloride, 0.5%, and phenylephrine hydrochloride, 0.5%, 45 and 30 minutes before examination. Indirect ophthalmoscopy was performed. If needed, topical anesthesia, an eyelid speculum, and scleral indentation were used to visualize the border between vascularized and nonvascularized retina. Infants who did not tolerate screening for ROP as scheduled were examined as soon as recommended by the neonatologist.
After birth, all infants in Sweden who fulfilled the study criteria (<27 weeks' gestation at birth) were immediately reported to the study coordinator and the responsible ophthalmologist at the infant's hospital. The tracking of infants and mothers was facilitated by the Swedish system of personal identification numbers given at birth. When infants were transferred to another hospital, the new ophthalmologist was contacted by fax or telephone to avoid delays in the screening program. After ROP screening was completed, the screening protocol of each infant was sent to the Department of Ophthalmology, University Hospital, Uppsala, where the findings were entered into a database, the Perinatal Quality Register, which also contained obstetric and neonatal data. All entries regarding ophthalmologic findings were checked against the protocols before the database was closed. The study was approved by the Ethics Committee, Medical Faculty, Lund University.
Statistical analyses were performed using a commercially available software program (Gauss; Aptech Systems Inc, Maple Valley, Washington). Odds ratios (ORs) with 95% confidence intervals were calculated using multiple logistic regression analyses. The GA was analyzed using class variables or, if specified, was entered into the models as a linear continuous variable. If specified, adjustments were made for BW (continuous variable). The Spearman rank correlation coefficient was determined when evaluating the correlation between rank-transformed data (eg, assessment of the correlation between severity of ROP and GA at birth).
Background data of the total cohort of 707 live-born infants in Sweden between April 1, 2004, and March 31, 2007, are given in Table 1. No infant had a major abnormality. One infant with Down syndrome was excluded. Two hundred infants died before the first ROP examination, which left a study cohort screened for ROP of 506 infants (229 girls [45.3%] and 277 boys [54.7%]; 410 single births [81.0%] and 96 multiple births [19.0%]). At least 1 eye had ROP in 368 infants (72.7%). The GAs at birth and the BWs of the total cohort and of those with and without ROP are given in Table 2.
Of the 506 infants, 192 (37.9%) developed mild ROP (stages 1-2) and 176 (34.8%) developed severe ROP (stages 3-5). Ninety-nine infants (19.6%) were treated. The incidences of maximum stages of ROP in relation to GA at birth are given in Table 3. Among infants with ROP, the severity of ROP showed a negative correlation with GA at birth (Spearman ρ = −0.25, P < .001).
Three logistic regression analyses were performed. In the first, GA was entered as a class variable, with 26 completed weeks as a reference. In the second analysis, GA was entered as a continuous variable in a linear model. In the third analysis, GA and infant BW were entered as continuous variables in a multiple linear model. In the Figure, the class variable ORs were almost the same as the estimates obtained from the linear univariate model using GA as a continuous variable. The series of ORs for severe ROP were 16, 7, 4, and 2 for the classes during weeks 22, 23, 24, and 25, respectively, compared with 26 weeks. This corresponds to an OR of 0.51 for each additional gestational week obtained from the model using GA as a continuous variable (Table 4).
When adjustment was made for infant BW, the association between GA at birth and severe ROP declined but remained statistically significant. Both GA at birth and infant BW were independently associated with severe ROP, but GA at birth was a stronger predictor of severe ROP than was infant BW (P < .001 and P = .004, respectively). Results of the univariate and multivariate regression analyses are given in Table 4. The ORs for any ROP did not differ substantially from the estimates obtained for severe ROP.
The present national, population-based study of extremely premature infants with a GA less than 27 weeks at birth shows a 73% incidence of ROP. The incidence was reduced from 100% in the 5 infants born at 22 weeks' gestation to 56% in those born at 26 completed weeks. In addition, the risk of ROP declined by 50% for each week of GA at birth in the cohort.
In the late 1980s, a few strictly population-based studies2,11,12 on the incidence of ROP were conducted. Since then, survival rates have significantly improved, and we have a new population of extremely premature infants.13,14 Despite improvements in the treatment of neonates and the increase in survival rates, no significant reductions in major morbidities, such as ROP, visual impairment, and cognitive disabilities, have been achieved.3,6,15,16 We, therefore, believe that the present epidemiologic study on the incidence of ROP in infants with a GA at birth of less than 27 weeks is of particular interest.
This study has several advantages. It is population based, with data on all infants born in Sweden. The infants are tracked with the help of a personal identification number given at birth. Another important advantage is that ROP is a primary outcome and that the ophthalmologic part of the study was designed and organized by ophthalmologists, who were also responsible for all data collection and input during the study. As in all national studies, its weakness is that many ophthalmologists examined the infants. We reduced this problem by giving standardized instructions initially and by repeating the information on classification and diagnoses throughout the study.
During the last decade, only a few national, population- based studies17- 21 have been published on the incidence of ROP with a sufficiently large number of infants with the lowest GAs at birth (22-24 weeks). The present prospective and national, population-based study covers 3 years and includes an even higher proportion of infants born in the earliest weeks of gestation (Table 5).
Comparison with other population-based studies on the incidences of ROP during the last 10 years is difficult because of differences in the designs and the inclusion criteria. As noted by Good,22 many studies have not included ROP as a primary outcome. Variations in survival rates and inclusion criteria regarding GA at birth and BW have also been present. Moreover, differences in screening routines and criteria for treatment of ROP may also affect the incidence. Similar to the population-based studies referred to previously herein,17- 21 we included all infants who underwent ROP screening in the study cohort, regardless of the number of eye examinations. Screening efficacy and its effect on the incidence of ROP will be reported in a forthcoming article. In view of the previous findings, it is not surprising that a recent national Norwegian population-based study18 on extremely premature infants revealed a lower incidence of ROP, and Belgian,17 Australian/New Zealand,19 Austrian,20 and Finnish21 population-based studies noted lower incidences of severe ROP than did the present study (Table 5). The higher incidence of ROP in the present study may be because of the higher proportion of infants born in the earliest weeks of gestation (ie, 11.5% of infants in weeks 22-23 vs 0%-6% in the other studies) (Table 5). These extremely premature infants, who previously did not survive, are probably especially vulnerable and prone to develop complications such as ROP. This agrees with the findings of a Swedish study3 on a change in the distribution of ROP in 2 consecutive population-based studies in the Stockholm area, with an increased risk of ROP in the most premature infants.
Comparisons with findings from other studies should also consider the stage and severity of ROP. The previously mentioned studies included neither a prospective ROP screening designed and initiated by ophthalmologists nor ROP as a primary outcome. This may have resulted in less efficient data recordings of the ROP screening and fewer reports of milder stages of ROP, which would lower the total incidence of ROP. Indeed, it seems more relevant from the clinical viewpoint to compare incidences of severe ROP, which frequently requires treatment.
In accord with previous population-based studies,2,11,12,18,19,21 the incidence of ROP was strongly associated with GA at birth in this extremely prematurely born cohort. It has been suggested that the most premature infants, born in gestational weeks 22 to 24, differ from the more “mature” infants regarding morbidity and outcome, which was also indicated in recent studies that include aspects of ROP.23,24 The findings on the incidences of ROP in different gestational weeks in the Norwegian and Finnish studies confirmed this view (Table 5).18,21 The Austrian study, however, reported a gradual reduction in the incidence of severe ROP with an increase in GA at birth.20 In the Australian/New Zealand study, Darlow et al19 observed a log-linear trend in the effect of GA on infants born after 25 to 28 weeks' gestation, but did not assess GA below 25 weeks separately. The present study shows a log-linear relationship between severe ROP and GA at birth even in gestational weeks 22 to 26. In contrast with the view that the most premature infants differ from the more mature ones, these findings, together with those of Darlow et al,19 suggest that the log-linear relationship between GA at birth and the risk of severe ROP is true of the entire extremely premature period, which includes all infants born before 29 gestational weeks.
We found that the association of GA at birth with ROP remained the most significant even after adjustment for BW. Gestational age was used as a cutoff variable (ie, a criterion for inclusion) in the present study rather than BW, as in many other studies and that of Tommiska et al.21 The inclusion criteria, such as GA at birth and BW, may also affect the incidence of ROP. For example, using BW as a cutoff variable can result in an overrepresentation of growth-retarded infants, whereas premature infants with a high BW may be excluded. The effects of BW, growth retardation, and early growth will be separately analyzed in subsequent articles.
In conclusion, improvements in neonatal care have increased the survival rates of a new population of extremely preterm infants. The present national, population-based study shows that such infants are at high risk for ROP and that the risk is reduced by 50% for each increase in the week of GA at birth.
Correspondence: Gerd E. Holmström, MD, PhD, Department of Ophthalmology, University Hospital, 75895 Uppsala, Sweden (Gerd.email@example.com).
Submitted for Publication: December 23, 2008; final revision received February 4, 2009; accepted February 6, 2009.
Author Contributions: Drs Källen and Holmström had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Birgit and Sven Håkan Olsson Foundation, the Evy and Gunnar Sandberg Foundation, Kronprinsessan Margarethas Arbetsnämnd för synskadade, the Norwegian Association of the Blind and Partially Sighted, Stiftelsen Solstickan, and the Swedish Association of the Visually Impaired.
Additional Contributions: We acknowledge the contributions of the following members of the EXPRESS (Extremely Preterm Infants Study in Sweden) Group: M. Blennow, U. Ewald, V. Fellman, T. Fritz, L. Hellström-Westas, P.-Å. Holmgren, G. Holmström, A. Jeppsson, K. Källen, H. Lagercrantz, R. Laurini, E. Lindberg, A. Lundqvist, K. Maršál (principal investigator), T. Nilstun, S. Nordén Lindeberg, M. Norman, E. Olhager, P. Otterblad Olausson, I. Östlund, F. Serenius, M. Simic, G. Sjörs, L. Stigson, K. Stjernqvist, B. Strömberg, M. Wennergren, and M. Westgren. Regional Ophthalmologic Coordinators: A. Hellström, G. Holmström, P. Jakobsson, K. Johansson, K. Tornqvist, and A. Wallin. Ophthalmologic co-investigators: S. Andersson, S. Andreasson, B. Arif, I. Axelsson, L. Berglin, I. Berndtson, E. Bonthron, B. Carlsson, P. Carlström, G. Cernerud, M. Cheraghchi, A. Collath, S. Crafoord, K. Dimitriou, A. Felix, Y. Friberg Riad, Å. Fridman, K. Fromm, L. Gustad, N. Hagwall, M. Hammarbäck, L. J. Hansson, K. Hellgren, L. Hilmertz, K. Holm, M. Hök Wikstrand, E. Hörnblad, A.-L. Hård, G. Ilic Sundström, N. Jadidi, G. Jakobsson, S. Jarkman, K. Johansson, P. Karlsson, L. Kjellberg, D. Kjellgren, T. Krebser, M. Kuusik, A. Kvanta, E. Larsson, U. Liden, C. Liljedahl, G. Lindgren, G. Lindgärde, C. Lundberg, I. Lundqvist, O. Lundqvist, Ö. Lundström, T. Lunt, E. Maly, T. Marczuk, Å. Nilsson, B. Nordin, G. Olivestedt, G. Olsson-Lalor, W. Owczarska, G. Pehrsson-Palmqvist, J. Rudebjer, A. Sellman, L. Sjödell, B. Sjödin-Backlund, B. Steen, E. Streman, B. Sunnqvist, S. Svedenhag, I. Taylor, O. Textorius, K. Teär-Fahnehjelm, I. Theocharis, Z. Tomic, O. Wennhall, Å. Wieslander, A. Wrigstad, A. Östberg, H. Åkerblom, G. Åkerskog, and S. Åström.